5182
NORBERT MULLERAND [CONTRIBUTION FROM
THE PHYSICAL
JEROME
GOLDENSON
Vol. 78
RESEARCH DIVISION,CHEMICAL WARFARE LABORATORIES]
Rapid Analysis of Reaction Mixtures by Nuclear Magnetic Resonance Spectroscopy1 BY NORBERT ~ T U L L E R ’ ~AND
JEROME
GOLDENSON
RECEIVED JUNE 14, 1956 Nuclear magnetic resonance spectroscopy is found to be a highly effective tool for the analysis of certain reaction mixtures involving phosphorus compounds. This is shown by the application of this technique to the isomerization of the insecticide P-ethylmercaptoethyl diethyl thionophosphate (Systox) . Results indicate that the rate law for the isomerization is intermediate between zero and first order, probably indicating a catalytic effect.
Introduction The field of high-resolution nuclear magnetic resonance (NMR) spectroscopy has grown rapidly since 1949, when Knight2 published his discovery of “chemical shifts” of the resonance of P31nuclei. He found that a t a fixed radio frequency the precise value of the magnetic field a t which resonance occurs depends not only on the nature of the nuclei involved, but also on their chemical environments. In several more recent the results of chemical shift measurements of over two hundred phosphorus compounds are presented. These observations show that NMR spectroscopy is potentially a very valuable method of quantitative analysis of mixtures of phosphorus compounds. A mixture which is to be analyzed by high-resolution NAIR must be a liquid, and the components must have chemical shifts differing by several chemical shift units. The unit usually employed is 6 =
(\-)
fir
I
x
106
where H , and H, are the resonance field values of the sample studied, and of a reference compound (usually HSP04), respectively. The strength of each P31 resonance should be proportional to the number of nuclei contributing to it. Thus relative concentrations may be deduced from relative signal strengths, without a need for absolute intensity measurements. Since the sample is not destroyed during the analysis, the NMR method is particularly useful in following the changes of concentrations in reacting mixtures without disturbing the course of the reaction. This paper reports such an application to the study of the kinetics of the isomerization of the insecticide p-ethylmercaptoethyl diethyl thionophosphate (“Systox,” I) t o the thiol isomer (“Isosystox,” 11).
I
I1
Each of these molecules exhibits a single P31 resonance, somewhat broadened by indirect spin(1) (a) T h i s article was presented as a paper a t t h e Pittsburgh Conference on Analytical Chemistry a n d Applied Spectroscopy, Febr u a r y 27. 1‘32T,. ( b ) Department of Chemistry, Purdue University, I a f a y e t t e , 1ndi;rna. ( 2 ) W. L). Knight, P h y s . K r u . , 76, 12BY (1049). ( 3 ) H. S. Gutoivsky a n d D. M‘. McCall, J . Cheix. I’hys , 21, 271) (1953). (4) S . Muller, P . C . Lauterbur and J , Guldenson, ’THIS J O U R N A L , 78, 3557 (1933). ( 5 ) J . R. Van Wazer, C . F. Callis, J . S . Shoolery a n d K . C . Jones, ihili., 78, in press (19.56).
spin coupling between the phosphorus nucleus and nearby protons. Analysis is readily possible, since the two signals are obtained a t values of the magnetic field differing by about forty parts per million, which is well above the resolving power of the instrument. Experimental Measurements were made with a commercial high-resolution NMR spectrometer obtained from Varian Associates, Palo Alto, California. The fixed radio frequency used was 17 mc./sec., with magnetic fields near 9850 gauss in the twelveinch electromagnet. The amplified magnetic resonance signal was displayed as needed either on an oscilloscope or on a Sanborn industrial recorder. Differences in the extent of indirect spin-spin interactions, and in relaxation times, make it unlikely in general that different P31 resonance signals will have the same width. Therefore, the areas rather than the heights of resonance peaks should be used as a measure of signal strength. We found that this could most conveniently be done by recording the spectrum and cutting out the individual peaks, and then weighing them on a n analytical balance. Since both materials have the same molecular formula, the mole or weight fraction of isomer I in the mixture should be given simply by the ratio of the weight of its own peak t o the total weight of the peaks. T o minimize random errors caused by transient electrical effects and human fallibility in locating the base line, six or more recordings were always taken and the peaks corresponding to each component weighed together. Values of the mole fraction of isomer I were reproducible t o within one or two hundredths. Systematic errors might arise either from non-linearity in the response of the ampliiier and recorder, or from a possible variation of the relative relaxation time ( T I )for the two isomers. To show that such errors were unimportant, measurements were made on samples prepared from known weights of the purest available Systox and Isosystox. The mole fractions found agreed within two or three hundredths with the calculated values. The Systox used in these studies was prepared in the organic branch of these laboratories, and distilled through a falling-film molecular still a t 58” and mm. Its refractive index was 1.4865. The Pal NMR spcctrum showed initially a small amount, perhaps 3% of Isosystox, and no detectable trace of other impurities. Contaminants not containing phosphorus would not be expected t o affect the P3I N N R spectrum, though they might influence the course of the isomerization reaction. Each sample consisted of about 1.5 mi. of the material sealed under vacuum in a Pyrex tube of 8 mm. outside diameter. The isomerization was followed a t 100” and a t 115” by immersing the samples in constant temperature baths for appropriate time intervals. After each heating period the samples were cooled t o quench the reaction, and the concentrations determined by NMR analysis.
Results and Discussion The chemical shifts of Systox and iiieasured by a method based on that and Packard,‘j were 6 = -67.7 and spectively. The data were typical of tained in Table I.
Isosystox, of Arnold -25.9,
re-
those ob-
(6) J . T. Arnold a n d 31, E;. Packard, J . Cheln. Phys.. 19, 1608 ( I D , i 11.
Oct. 20, 1956
PURIFICATION OF
ANHYDROUS HYDROFLUORIC ACID TABLE I1
TABLE I ISOMERIZATION OF SYSTOX AT 115' Heating time, hr.
0.5 1.1 2.0 3.0 4.0 5.0 6.0
% Systox remaining
Sample 1
Sample 2
87% 77
87% 77 63 50 36 26 20
61
5183
Run Run Run Run
1, 100" 2, 100' 1, 115" 2, 115"
ki
ki
0.071
0.49 .59 .42 .49
.078 .25 .26
rapid equilibrium is first set up between the reagent and a small amount of catalyst, reacting to form an intermediate, the decomposition of which is relatively slow. In this scheme, kl is the recipro18 cal of the equilibrium constant for the first reacAnalysis of the data shows that the rate law for tion, while k 1 is the product of the rate constant for the reaction is intermediate between zero and first the decomposition of the intermediate and the cataorder, rather than simply first order, as had been lyst concentration. Although this equation may reported by Fukuto and Metcalf.' Their method also be derived on other assumptions, it seems likely involved the use of Systox labeled with radiophos- that a catalytic effect is involved. This idea rephorus; samples withdrawn from the reaction mix- ceives further support from the fact that our measture were analyzed by separating the isomers ured half-life of about 11 hours a t 100' is longer chromatographically and measuring the radioactiv- than the 8.8 hours calculated from Fukuto and Metcalf's data a t 9.5'. Their material was preity of each fraction. pared by a method which would not be expected to Our own data may be represented by a rate law yield very pure products. of the form NMR spectrometry is found to be a very convenklc - -dc =ient tool for analytical problems of the type endt kz c countered here. The fact that all samples are The best values for the k's were found by the sealed in glass throughout the study reduces the method of least squares, using two runs a t each of danger of working with these toxic materials, and the two temperatures. The unit of time used was the chances of having the samples contaminated. hours and the concentration, c, was expressed as Other advantages are the small amounts of matemole per cent. The numerical values of the K's are rial required, and the convenience with which data shown in Table I1 may be obtained. The accuracy of NMR anaThe equation is of a form discussed by Laidler,* lytical methods will undoubtedly be improved as which corresponds to a reaction scheme in which a the instruments are further developed, but it is al(7) T. R . Fukuto and R . L. Metcalf, THISJOURNAL,76, 5103 ready as good as or better than that of most rapid (1954). methods used in kinetic studies. 47 35 24
+
(8) K. J . Laidler, "Chemical Kinetics." McGraw-Hill Book Co., New York, N. Y . , 1950, p . 279.
[CONTRIBUTION FROM T H E
ARMYCHEMICAL CENTER,MD.
DEPARTMENT O F CHEMISTRY
O F ILLINOIS IXSTITUTE O F TECHNOLOGY]
The Purification and Specific Conductivity of Anhydrous Hydrofluoric Acid BY MERVINE. RUNNER, GEORGEBALOGAND MARTINKILPATRICK RECEIVED MAY18, 1956 Improvements in the methods for the purification of anhydrous hydrofluoric acid have yielded acid of much lower specific conductivity than reported previously.
The ion product for hydrofluoric acid was estimated by Kilpatrick and Luborskyl t o be 2 X a t 20'. On the basis of this value a specific conductance for pure anhydrous hydrofluoric acid of 7 X ohm-l ern.-' a t 20" was ca1culated.l This conductance was considerably lower than the value 1.4 X ohm-' cm.-l a t - 15" previously reported by Fredenhagen and Cadenbach. However, in practice, i t has not been experimentally feasible to obtain and handle hydrofluoric acid of purity even similar to that of Fredenhagen and Cadenbach.2 Kilpatrick and Luborsky3 used hy(1) M . Kilpatrick and F. E. Luborsky, THISJOURNAL, 76, 5865 (1954).
(2) K . Fredenhagen and G. Cadenbach, 2. u n o ~ g .allgem. Chem., 178, 289 (1929). (3) M. Kilpatrick and F. E. Luborsky, THISJOURNAL, 71, 577 (lY.53).
drofluoric acid that had a specific conductivity of 1 X lo-* ohm-' c m . - l a t 20" and more recently the conductivity measurements of Rogers, et a1.,4 were carried out in hydrofluoric acid of similar purity. Since techniques for fabricating equipment of Fluorothene or Kel-F plastic (trifluorochloroethylene polymer) had been developed a t this Laboratory,6 the construction of hydrofluoric acid purification equipment of plastic was undertaken. We have found that with this equipment, relatively large volumes of hydrofluoric acid of specific conductivity lower than the above value calculated by Kilpatrick and Luborskyl are obtainable by fractional distillation. (4) M. F. Rogers, J . L. Speirs, M. B. Panish and H. B. Thompson, ibid., 78, 936 (1956). ( 5 ) M. E. Runner and G. Balog, Anal. Chcm., 28, 1180 (1956).
